iTEft

A STUDY OF A DRY KILN WALL IMPLOSIONBy: CLAUDE J. DOUTY, P. E.

SUMMARYOn February 6, 1957, an outside wall of a new dry

kiln installation was severely damaged by an im-plosion. The cause of the implosion was not known.Investigations conducted at that time were inconclusive.This led to the setting up of a series of tests in order todetermine what factors contributed to the implosion.

Rapid condensation of the vapor in the kiln ap-peared to be responsible for the implosion. The prob-lem was to determine what caused the vapor to con-dense. There was nothing that appeared to be unusualabout the weather or the lumber charge. The weatherwas freezing and foggy, while the lumber charge wasfrozen and snow covered.

Three theories were advanced as a basis for thecause of the implosion. These theories were based oncold air, fog, and frozen lumber. Cold air and fog wereeliminated because calculations indicated that the vol-ume of the condensed vapor was less than the volumeof air, or fog, required to condense the vapor. Frozenlumber appeared to be a possible cause.

Recording instruments installed in the test kiln indi-cated that the low pressures developed during theearlier part of the drying cycle, and when the fansreversed. Controlled tests indicated that it is unlikelythat the implosion resulted from a fan reversal whilethe steam spray was on.

The kiln controller chart that was in use when theimplosion occurred shows that the warm-up rate wasslower than normal. This slow warm-up rate was dup-licated by heating a kiln of frozen lumber with thesteam spray on and the fans off. Atmospheric testsmade in the kiln indicated that a dangerous conditiondid exist.

It appears probable that the implosion resulted fromthe fans not being on at the start of the drying cycle.The steam spray increased the vapor content in thekilns and then when the fans started the vapor wasforced through the frozen lumber. The vapor condensedwith a resultant pressure drop which caused the wall tofail

At some unknown hour on the morning of Febru-ary 6, 1957, an implosion severely damaged an exteriorwall of a new dry kiln at the Emmett mill of the BoiseCascade Corporation. Investigations conducted at thetime of the implosion brought forth so many conflictingstatements that it was not considered advisable to drawany conclusions from them. This led to the setting upof tests in a kiln at the plant where the implosion oc-curred. The purpose of the tests was to determinewhat unusual conditions existed in the kiln, what con-ditions could cause an implosion, and how to preventanother one. The tests were conducted with the as-sistance of Mr. Ed Knight of the Western Pine Labora-tories.

There is evidence that implosions have occurredelsewhere. However, they have not been as severeand most of the observations showed damage limitedto the doors.

The dry kilns are of the cross-ventilation type (Fig.1) with the fans mounted overhead and on the output

Fig. 1 Kiln Cross Section Drawing

A mu nation sketch showing the ■enoral arrangement of the crowventilation typo dry kiln.

shafts of gearhead motors. Heat is provided by over-head steam coils and there is an overhead 3 inch sprayline. There are two rows of automatic vents and tworows of hand controlled vents. The building is con-structed with an insulated structural concrete roof and

Mr. Claude J. Douty:Mechanical Engineer forBoise Cascade Corpora-tion.Mr. Douty, born and raised

in Washington, graduatedfrom State College of Wash-ington. He has worked "Dia-mond Match Company," TheWood Products Laboratoryat the State College of Wash-ington, and Boise CascadeCorporation.

— 7 —

04,Figure 2. Angle View of Wall—The fan support beams can be seen

projecting through the wall. The broken pilasters are hay-dite blocks filled with reinforced concrete.

block walls. Each kiln is a single track unit, 104 feetlong, by 17 feet-4 inches wide, and 24 feet high.

Operation of the kilns requires a minimum amountof attention. Fan reversals are made automatically.The kiln controller regulates the heating coils, the sprayline, and the automatic vents. Hand controls are pro-vided in case of the failure of the controller, or forchanges in operational requirements not allowed for bythe controller.

Standard operating procedure at the time of the im-plosion was to close the hand vents and leave thehumidity control on full automatic. The controller thenclosed the automatic vents and opened the spray valve.Fan reversal was on a full automatic schedule with nozero reset at the beginning of the drying cycle.

The lumber charge, of 4/4 select, went into the kilnat 2 a.m. It was frozen and snow covered. The am-bient temperature at that time was recorded by thepower house at 27°F. At 8 a.m. the ambient tempera-ture was recorded at 10°F. and at 9 a.m. it was 12°F.

Figure 3. Side View of Wall—The hole in the wall was made by the`implosion. It measured approximately 2 1/2 ft. by 2 1/2 ft. The blocks-Were pulled into the kiln with enough force to knock several boards

off the top of the load.

The implosion was discovered between 8 a.m. and 9a m. (Fig. 2 & 3). At the time of discovery the weatherhad become foggy.

This is about all of the positive information thatwas available to investigators. A shift change at 7a.m. caused a second crew of operators to become in-volved. The men were nervous and their statementsfailed to shed any light on the problem.

There was only one logical explanation for theimplosion. It had to be the rapid condensation of watervapor within the kiln. Therefore, the first thing to dowas to stop using the spray during the early stages ofthe drying cycle. This change in operating procedurewas made immediately.

Figure 4. Safety Door—A spring loaded safety door was installedin each main door to relieve any low pressures that should develop.

The second precaution taken was the installationof spring-loaded safety doors. These doors (Fig. 4)were installed in the main kiln doors and were designedto open under a very low pressure differential.

The main question was "what caused the watervapor to condense." There were three possibiltiesavailable, and each of them was the basis of a theory.These were—fog, cold air, and frozen lumber.

According to a staff member of the Boise weatherstation, fog is always made up of tiny water droplets,regardless of temperature. We didn't know much aboutthe nature of fog, so it was decided to check into thisone a little further.

Cold air could be responsible. Some of the ventswould have to be open, but there is no record showingthat they were open or closed. However, several doorswere drawn in prior to the failure of the kiln wall. Weknow that at times the atmosphere in the kilns did ap-proach critical conditions while they were being oper-ated in a normal manner.

Frozen lumber had been dried before. In fact therewere times when the kilns had been charged with lum-ber that had been stored under worse conditions. How-ever, there was no evidence to single out this particularcharge as being different or unusual.

Fog was investigated theoretically by means of aheat balance based on values taken from the steamtables.

The worst possible condition that could be set up inthe kilns would be to have a temperature of 212°F. andthe air saturated. Under these conditions, the followinginformation can be taken directly from the steam tables.

1. Volume of one pound of saturated vapor=26.8 ft.°

2. Volume of one pound of saturated liquid=.01672 ft.°

3. Heat of evaporation of one pound of saturatedliquid=970.3 B.T.U.

4. Heat in one pound of saturated liquid=180.07B.T.U.

However, fog below an ambient temperature of 32°F.is a super-cooled liquid. The heat required to raise apound of fog from 0°F. is approximately 212 B.T.U.

The maximum amount of heat that could be trans-ferred to a pound of fog would be 212 B.T.U. using 0°F.as a base line. Any additional heat would start toevaporate the fog which in turn would reduce thevolumetric changes. If one pound of fog could absorb212 B.T.U., it would require approximately 4 1/2 poundsof fog to absorb the 970 B.T.U. required to condense apound of vapor. If this could be done there would bea volumetric reduction of approximately 26 cubic feet,and a pressure drop to correspond to this change.

The amount of air required to suspend 4 1-/-, poundsof fog should also be considered. To do this we willconsider another extreme condition. The maximumamount of water that a cubic foot of air could possiblyhold as fog would result from cooling saturated air at212°F. until all of the vapor had condensed with nofall out.

1. If one pound of saturated vapor at 212°F. oc-cupies 26.8 cubic feet, then one cubic foot of vaporwould weigh approximately .037 pounds.

2. If one cubic foot of air could hold .037 poundsof water in the form of fog, it would require approxi-mately 122 cubic feet of air to hold 4 1/, pounds of fog.

The Encyclopedia Britannica states that a cubicmeter of fog will hold approximately three grams ofwater. This is equivalent to approximately .00019pounds of water per cubic foot of fog.

Under ideal conditions it would require 122 cubicfeet of fog-laden air to enter the kiln in order to con-dense 26.8 cubic feet of vapor. This is an impossibilitysince the volume taken in would more than fill the voidleft by the condensing vapor. It, therefore, eliminatesfog as a possible cause of the implosion.

Cold air was eliminated because of its low heatcapacity. For example, the heat capacity of air is .24B.T.U./lb. °F. At 68°F. one pound of air will occupy13.3 cubic feet of space. At 212°F. it will occupy morespace, so the 13.3 cubic feet value is conservative forthis case. One pound of air raised from 0°F. to 212°F.will require approximately 51 B.T.U. To get the neces-sary 970.3 B.T.U. required to condense a pound of vaporwould require approximately 19 pounds of air occupy-ing a volume of approximately 253 cubic feet. This is

another impossibility because the volume of air requiredis greater than the void left by the condensing vapor.

A charge of 4/4 lumber may have 130,000 squarefeet of surface. If the surface temperature of the lumberis cold enough, it would make a beautiful condenser.Frozen lumber certainly could be a contributing factor.

To be absolutely sure of the cause of the implosionit would be necessary to set up a test kiln where avacuum could be produced at will under controlled con-ditions.

There was no question that we were playing witha powerful force. It would certainly be embarrassing,to say the least, to have one of the tests pull in, anotherwall. We decided to install recording instruments inthe test kiln and just record the normal kiln operationuntil enough information was obtained to justify anytest procedure.

Even though fog and cold air were not consideredresponsible for the implosion, we wanted to knoweverything we could about the workings of the drykiln. To achieve this we set up instruments to recordambient air temperature, wet and dry bulb readings,the opening and closing of the automatic and hand con-trolled vents, the operation of the spray valve, flow ofair in through, or out of, the vents, and the pressure dif-ferential between the inside and outside of the kiln. Onepressure point was located about four feet from the floorand beside the outside wall. The other two points weretaken above the first point with one on each side of afan and located on the fan center line. In addition tothis information, the kiln operators recorded the weatherconditions each hour for us. They also marked on thepressure chart when they changed the kiln and the runnumber.

All charts were on a 24 hour basis. The pressurerecorder could also be set for a 96 minute chart if de-sired. The reason for this was that the standard kilnchart is based on seven days. Disturbances recordedon such a chart would be difficult to distinguish be-cause the pen recording would be too compressed tobe readable.

After the first series of observations were com-pleted, the recorded data was analyzed. The importantfactors influencing the pressure changes in the kilnswere noted. These were found to be:

1. Fan reversals coincided with the developmentof the vacuums.

2. The vacuums were noted to occur in the earlyphases of the cycle. For this reason it was con-cluded that the condition of the lumber in thecharge had a bearing upon the development ofthe vacuums.

Intermediate tests were conducted without the useof the spray. Limited vacuums of one inch of waterpressure were obtained. Since the winter of 1957-58was very mild it was decided that the tests should becontinued for another winter.

In the fall of 1958 the recording instruments werestarted again. Cool weather in the middle of Novemberprovided a chance to do some testing without the dan-ger of extreme conditions.

At 7:30 a.m. on the 16th of November a charge of4/4 red fir common was loaded into the test kiln. This

-9-

lumber had been stored in temperatures ranging from24°F. to 40°F. About one-half of this time was belowthe freezing point. The lumber was subject to rain andsnow. The safety door was held closed by an arrange-ment of springs. This was done to increase the openingresistance.

The test procedure was to operate the fans for afew minutes to allow the atmosphere in the kiln to reachequilibrium. The spray was turned on ten seconds be-fore stopping the fans, and allowed to continue forvarying lengths of time. When the fans started up, thepressure recording pens would oscillate several timesbefore returning to zero pressure. The spring-loadedsafety door was opening and relieving the vacuum.The springs would then close the door and the vacuumwould build up again.

Pressure differentials of one inch of water wererecorded. Restraining the safety door while the fanswere started produced a vacuum of three inches ofwater. One inch of water is equal to a pressure of 5.2pounds per square foot. A pressure of three inches ofwater is equivalent to 15.6 pounds per square foot, or atotal force on the kiln wall of over 19 tons. It was alsoobserved that the duration of the reduced pressure wasa matter of only a few seconds.

The results of this test were very encouraging be-cause we had produced a partial vacuum under con-trolled conditions. However, additional informationwas necessary to correlate the conditions within the kilnto the intensity of the vacuum produced.

A lumber charge consisting of 4/4 half & half pinecommon was placed in the test kiln. The temperatureof the lumber was 46°F. The test procedure consistedof spraying the kiln for 15 seconds before stopping thefans. The spray was continued for varying lengths oftime during the reversal cycle. Wet bulb and dry bulbreadings were taken when the fans started. Lumbertemperature readings were taken immediately after-wards. The safety door was blocked for this test.

The final run was made approximately two hoursafter the kiln was charged. The spray was on for thefull time the fans were reversing (2 minutes, 15 sec-onds). Wet bulb temperature was 135°F. and dry bulbtemperature was 153°F. Lumber temperature was120°F. and the vacuum developed was 3.0 inches ofwater.

Test data indicated that the vacuum produced dur-ing the fan reversal was a function of a length of timethe spray was on. When the spray was on for one andone-half minutes the vacuum produced ranged fromabout one-third to one-half of the vacuum producedwhen the spray was on for two minutes and fifteen sec-onds.

Tests made by the Western Pine Laboratory haverecorded a temperature drop of 22°F. across a 9 footload of Ponderosa Pine sapwood lumber during theearly portion of the kiln drying period. We recordeda temperature drop of 19°F. through frozen spruce inour Emmett kilns. Multiple passes through the lumberdoesn't appear to increase the intensity of the vacuumproduced, or to lower the air temperature. Each cycleof the air includes a pass through two banks of finnedpipe with a temperature of approximately 335°F. Thenet result is a gain in air temperature. Of course, the

temperature of the air entering the lumber would havea bearing on the temperature drop.

Our testing entered a new phase at this point. Webegan the investigation of atmospheric changes thattake place in a kiln when the spray is on and the fansare off.

After charging the kiln with frozen lumber, the fanswere turned on for a few minutes to establish equili-brium conditions. The spray was turned on and thenthe fans were stopped. The spray was allowed to con-tinue for about 3 1/, minutes. At the end of the sprayingtime the atmosphere under the kiln roof had a dry bulbtemperature of 186°F. and a wet bulb temperature of128°F. The atmosphere six feet from the floor had adry bulb reading of 100°F. and a wet bulb reading of80°F.

The dew point for the atmosphere under the roofwas approximately 123°F. This means that a tempera-ture drop of 63°F. would be required before condensingvapor could create a vacuum. An additional drop of8°F. would be required to develop the 50 p.s.f. vacuumthat the walls were designed to withstand. At the 6 ft.level we had a dew point of 73°F. A temperature dropof 27°F. would be required to create any vacuum. Atotal temperature drop in excess of 68°F. would be re-quired to endanger the walls.

These conditions were recorded after the spray hadbeen on for approximately 31 2 minutes. When the im-plosion occurred the fans were only off for 1 1/0 minutes,and they coasted for all of that. Possibly 1%z to •/4 ofa minute was available when the fans were turningvery slowly, for vapor to accumulate in the kiln. It ap-pears probable that temperature drops in excess of22°F. do occur. The mixing of the atmosphere by thefans may also produce a more dangerous conditionthan exists at the instant the fans start.

Three variables appear to control the intensity ofa vacuum—the relative humidity and temperature ofthe kiln atmosphere, and the surface temperature of thelumber charge. In some cases lumber species mayhave a bearing, but if the lumber is coated with ice andsnow the species is of less importance. Higher relativehumidity and temperature require less of a temperaturedrop through the lumber to create a vacuum. Lowerlumber surface temperatures likewise can producevacuums with lower atmospheric temperature and rela-tive humidity.

The amount of steam injected into a kiln is constantfor each fan reversal period. For normal operation ofthe kilns, the only uncontrolled variable is the tempera-ture of the lumber charge. Prior to the implosion lum-ber charges had been subjected to temperatures of20°F. to 25°F. below zero. For some unknown reasonthese more severe conditions didn't destroy any walls.

It can not be stated definitely that the wall failuredidn't occur during normal operating conditions. How-ever, when all of the available information is consid-ered, it appears probable that the implosion was notthe result of a fan reversal occurring while the sprayvalve was open.

This cast doubts upon the last of the three theoriesthat were advanced at the time of the implosion. It stillleaves the question of how the critical conditions weredeveloped in the kiln.

— 10 —

The kiln controller chart (Fig. 5) that was in use atthe time of the implosion is rather confusing. There isone thing though that seems to have been overlooked.That is the indication of a very slow warm-up rate forthe first four hours that the charge was in the kiln. Weknow that if the spray is on and the fans are operatingthe warm-up rate is rather good, even with frozen lum-

The kiln controller chart madeat the time of the implosion.There are indications that theaction of the pens was erratic.The diagonal lines at the lowerleft side indicate a slow warmup rate. This is not normal, evenwith frozen lumber, if the fansstart at the beginning of thedrying cycle.

ber. There is a possibility that the kiln operator set thecontroller but that the fans failed to start.

Another test was set up to check this theory. Afrozen charge of 4/4 red fir common was placed in akiln. The fans were left off while the spray and theheaters were turned on. The surface temperature of thelumber was 24°F. when it entered the kiln. The kilncontroller appeared to require calibration, so wet anddry bulb thermometers were placed in the kiln at the6 ft. level.

After two hours of spraying the following condi-tions were recorded. Under the kiln roof, the dry bulbtemperature was 232°F., and the wet bulb temperaturewas 196°F. At the 6 ft. level, the atmosphere was satu-rated at 62°F. The lumber surface temperature was34°F. at a point about 10 ft. above the floor. Snow wason the lumber at the lower levels. After four hours ofspraying the following conditions were recorded: Underthe kiln roof the dry bulb temperature was 239°F. and

the wet bulb temperature was 202°F. There was nochange at the 6 ft. level. The lumber surface tempera-ture was 35°F. at the 10 ft. level and there was stillsome snow between the courses at the lower levels.

The dew point for the atmosphere under the roofwas approximately 201°F. A temperature drop of 38°F.would be required to start forming a vacuum. An ad-ditional drop of 4°F. could create a vacuum exceedingthe design strength of the wall. Under these conditionsthere exists a maximum temperature difference of 198°F.between the kiln atmosphere and the lumber. At theend of the four hour spray the top half of the kiln wasfilled with steam. This fact coupled with the 198°F.temperature difference between the lumber and theupper atmosphere created a more dangerous conditionthan anything we were able to produce during normaloperation. Actually cond i tions nearly as dangerousprobably existed in a matter of minutes after the spraystarted, since a 3 1V. minute spray produced a wet bulbreading of 128°F. and a dry bulb of 186°F.

After this four hour spray cycle the vents and thedoor were opened. The fans were started one at a timeand then the kiln was closed up again. The controllerchart (Fig. 6) shows the same basic warm-up patternthat appears on the chart made when the wall failed.

The atmospheric conditions that developed in thekiln during the four hour spray, and the similar chartrecord indicate that the implosion probably resultedfrom an operational error or a mechanical malfunction.This also explains the unusual conditions that existedin the kiln at the time of the implosion.

The safety doors limited a three inch vacuum toone inch. They are probably capable of doing muchbetter than that, however, we didn't want to exceeda three inch vacuum in our tests.

Door failures were occurring with the 1 1/. minutefan reversal period. A series of fuse failures in the fanmotors caused the reversal period to be lengthened to2 1/4 minutes. It is possible that the use of the spraycould cause a wall failure in normal operation of thekilns now where it didn't appear to before.

There are a few methods that could be employedto help keep dangerous atmospheric conditions fromdeveloping in the kilns while using the spray for thewarm-up period. Among these are: the installation oftimers to reverse the fans one at a time, the lengtheningof the fan operating periods while the lumber is cold,and the installation of solenoid valves in the sprayvalve air control lines. These solenoid valves wouldbe wired to the fan motor leads so that any time thefans were off the spray valves would be closed. Thelast method is the one to be preferred because it wouldkeep the steam out of the kilns when the fans are off.

Fig. 6 Kiln ChartWarming up a charge of frozenlumber without the use of thefans produced the same diagonallines at the start of the cycle.The controller was recalibratedafter this test.